My 2016 retrospective in which I talk about projects I worked on, pushed off and other things.
I spent some time in 2016 passing one of my big projects off:
PyVideo: In January 2016, I announced I was pretty done and wanted to end the site. I didn't have anyone to hand it off to, so that was going to be that. Sheila and I talked about that for a bit and then decided to export all the data into a new project and spend some quality time on improving the data and building infrastructure so a community could maintain the data.
Then Paul showed up with a cadre of like-minded people and built a new statically-generated PyVideo using the data Sheila and I had been curating. Sheila and I worked with him and it was a smooth transition. The site is doing a lot better now and has a larger community of people maintaining it. That's fantastic!
In October 2016, the Python Software Foundation awarded Sheila and I a Community Service Award for our work on PyVideo. I'm really proud of this even though I wasn't as successful as I wanted to be. There's a blog post covering PyVideo, the handoff and the award.
At this stage, I don't have many projects on my plate anymore. I continued working on existing projects:
- bleach: I continued to clean up issues and pull requests, fix documentation to clarify the scope of the project and worked on code changes for 1.4.3 and 1.5 releases. Started the work to switch to html5lib 0.99999999 (8 9s).
- Standup: Mozilla ended Persona which was the identity management system Standup used. Further, the codebase was pretty rough and it was hard to administrate the site. So Paul and I decided to rewrite Standup as a Django app and switch to Github auth. There are still a lot of issues, but the base is theoretically better and the site can keep operating even after Persona has been shut down.
- Dennis: I spent some time fiddling with translation pipelines. It's been tricky. I haven't worked out an architecture I like, yet. I don't use Dennis anymore and I don't have any help on it, so I think I may look to pass this project off or end it in 2017. If you're interested in it, let me know. I interpret silence as you're not interested and don't use it.
I started some new projects:
Everett: Everett is the result of me being unhappy with the landscape of Python configuration libraries. It started out as an experiment, but I'm now using it in a few projects and I'm pretty happy with it.
In 2017, I'll probably do more to make it a real project and write some posts on architectural decisions.
Antenna: At Mozilla, I'm currently on the team working on Socorro which is our system for crash report collection, processing and analysis for Firefox and other Mozilla products. In 2016, I embarked on extracting and rewriting the collector portion of the ingestion pipeline.
I spent some quality time with AWS, AWS Lambda, docker, Python 3, Falcon, gevent and boto3. I also developed some things that are being spun out into other projects: Everett, a logging mock, a boto3 S3 mock and a metrics infrastructure.
Generally 2016 was ok, but felt slower than previous years. I didn't feel like I got much accomplished. I switched from Input (now decommissioned) to MDN then to Socorro so I spent a large portion of the year onboarding onto big, complex projects.
My manager says this is all expected and that I really did accomplish a lot, but the nature of what I accomplished was implicit and not explicit and so harder to itemize.
I know I got some things done:
I wrote up onboarding docs for MDN developers.
I cleared the MDN security bug backlog.
I wrote Socorro onboarding documentation and cleaned up some of the project infrastructure documentation. There's a ton more work to do here, but it's probably better.
I rewrote the Socorro collector component. I read through the existing code, wrote out a spec, experimented with some architectures, settled on one, implemented it, wrote tests, wrote system tests, coordinated other peoples' efforts and worked on a migration plan.
I actually did more work than that--I wrote two partial implementations with two different methodologies and then started a third ground-up rewrite in September which is the one we're going with now.
When writing the new collector (called Antenna), I ended up writing a configuration library (Everett) to simplify the architecture.
I started a project to dockerize Socorro to make it easier to run a prod-like environment on my local machine. I have a collector and a crashmover. I experimented with how to structure it so that it's convenient for use in development. This needs to be finished (processor, middleware, webapp, crontabber), but I think I've settled on a good approach.
I assisted in upgrading Socorro from Python 2.7.5 to 2.7.11 which turned out to be super hard because of how our infrastructure was set up and the bash scripts we use to build and delpoy things are "interesting".
I took over the Socorro stage submitter which was having a lot of problems and kept falling down which disrupted development and testing. After taking it over, I cleaned up the node, reworked how it runs and wrote docs so others could fix it when it went down. (Though it hasn't gone down since.)
I worked with Paul to rewrite Standup as a Django app making it easier to maintain. We sill have a ways to go. We spent a lot of time dealing with authentication and identity issues. We switched from Persona to Github auth which created a lot of problems. Now we're switching from Github auth to Mozilla's hosted Auth0 which is creating a new set of problems. I think I'm going to have to fix everyone's accounts by hand at this point.
I spent a lot of time rewriting our Auth0 implementation so that it was easy to extract into a Django library. If that's useful to other Mozilla sites and I get some help with maintaining it, I'll split it out. Otherwise it'll continue on as an overengineered auth system in Standup.
I switched to instant coffee for 6 months. I don't think I'm going to do that again, but it was interesting.
Challenges for the year:
- Things feel slower than when I was working on Input.
- Onboarding is hard. Joining teams with established development processes, culture and lots of history is hard. Coordinating with other busy people is hard.
- Carving out time to do deep dives into big issues on side projects like bleach and html5lib 0.99999999 (8 9s) support has been difficult. I've always had this problem--it's nothing new, but it's still tough.
- Dropping my FSF membership in June. I probably should have done this a couple of years ago. Lots of reasons. Feeling much better about things now that I've dropped my membership.
In 2017, I want to:
- Clean up my office. It's a mess (still).
- Make awesome stickers for one of the things I work on.
That's 2016 in a nutshell!